The proposed research will examine how retarded children can be stimulated to use the language skills they have already acquired. There is theoretical and empirical support that frequent elicitation and support of mother-child conversation is associated with greater child language development in nonretarded children. However, retarded children have been described as more passive in verbal interactions than are their linguistically matched nonretarded peers. Studies are designed to: (a) evaluate the relative effectiveness of conversational recruiting strategies in eliciting the conversational participation of mentally retarded children who are in the one-word stage of language learning; (b) test the immediate effect of and general association between certain characteristics of the retarded children's speech and mother's use of these recruiting strategies; (c) determine whether individual child and mother differences seen during interaction predict the rate of retarded children's early language development; and (d) determine whether retarded children's conversational participation can be manipulated by assuming a prescribed interaction style. Two similar samples are proposed for the project. Each sample will be composed of 30 moderately to mildly retarded preschoolers who are in the first stage of language learning and their mothers. Subjects will be selected from all SES levels. The first sample will engage in six procedures. Mother-child pairs will engage in videotaped free- play (1) and familiar book reading (2) sessions for subsequent coding. Children's initial cognitive (Bayley MDI (3)) and language levels (SICD (4) and experimenter-child free-speech sample (5)) will be assessed. Mothers will watch games (routines) that they play with their children (6). One year later, the two assessments of the children's language level will be repeated. The second sample will also engage in mother-child free-play sessions. Additionally, the children in the second sample will interact with two adults. One of the adults will assume a "conversation- eliciting" interaction style. The other will assume a "neutral" interaction style. The analysis strategies are recently developed sequential analysis procedures, repeated measures ANOVAs, simple and multiple regression techniques, and a binomial test. Autocorrelation will be controlled when carrying out the sequential analyses.